So the THGC moderator suddenly woke up and decided to remove my thred: "Taynalator SUCKY!!!"
I can't blame them for doing that, because the headline of my thred looked like a flame. The actual content however was no such thing. I'm just tired of headlines like "P4 sucks", "AMD Athlon XP sucks" and so on. Normally I don't bother, because it is evident that whoever is posting is doing so in order to start a flame war. With Taylanator however, I felt that he was being serios about his claims, as if he actually believed in the nonsense he spewed. Also I wanted to let him know that headlines like his "P4 3010MHZ SUCKY!!!" is annoying and stupid.
OK, but I still want my message through to get through to him, so here is a repost of my original posting:
@Taynalator:
As it is simply too overwhelming to respond to every little cry you’ve made over the last couple of days, I think I’ll just make some general comments and then take it from here. To be fair, I have to admit that it seems you are becoming more reasonable as time goes by. You have probably discovered that your bullish/stupid style won’t do you any good in making arguments. Nonetheless I still feel a need to set things straight.
1.The test was made primarily to show the overclockability of the Northwood. After the die-shrink it makes sense to examine its overclocking capabilities.
2.The other CPU’s listed in the benchmarks have just been included as a reference. They’ve not been made especially for this article but were all made some time ago. Do not regard this as an insult towards AMD (and you). It’s very time-consuming to do these kinds of tests, so for now the THG people have chosen to focus on the new 0.13 micron CPU offering from Intel. Probably later we will see something similar done for the AMD Athlon/Throughbred.
3.As I have already said some time ago, I’ve noticed that it seems to be very important for some test-sites to be quickly out with their reviews. That’s somewhat understandable, and the penalty seems to be bad editing at times. That’s not good if the article is trying to explain technically difficult stuff, but when it comes to benchmarks, it really doesn’t matter that much. If you use a little common sense you can draw your own conclusions in each benchmark. As long as they don’t fiddle with the results, I’m mostly happy.
4.There IS a problem with the German to English translation, but it is nothing major. I have visited the German site and found that the translation is actually adequate (being a Dane, I also “master” the German language), but the German site had two comments that are left out in the English translation. In the SiSoft Sandra benchmark the following comment can be found on the German site:
<i>”Der Pentium 4/3000 liegt bei allen Benchmarks vorn. Dennoch sind die Ergebnisse nur unter Vorbehalt zu verwenden. ”</i>
That could have been said a bit more precisely, but they are effectively saying that the synthetic memory benchmark is not reflecting the actual performance of a system. The translator, being a clever person, couldn’t match the results with the comment and probably simply left it out. Instead the comment should have been something like this: <i> The overclocked PC2100 DDR RAM on the P4/3000A system is not able to match the bandwidth of a standard RDRAM based system.</i>
The other comment can be found in the Cinema 4D bench:
<i>”Beim 3D-Rendering unter Cinema 4D zeigt sich, dass AMD und Intel sich in der Führungsposition abwechseln.”</i>
At best, this is a totally irrelevant comment, which the translator wisely left out. The P4/3000A is taking the lead in both tests with a significant margin. What the editor is maybe trying to say is that the blue(P4) and green(Athlon XP) bars interchange in the Shading (Cinema 4D) benchmark. Totally irrelevant.
5.As far as judging the performance of the P4 as “PITIFULL”, I think you are completely wrong. The P4/3000A displays an almost linear performance/clockspeed relationship in the SAME benchmarks as the AMD Athlon XP. In the following table, I’ve listed what a 25% increase in clockspeed does for the Athlon XP (1333MHz/XP1500+ to 1660MHz/XP2000+) and what a 50% increase in clockspeed does for the P4 (P4/2000A to P4/3000A).
Benchmark:..............................Athlon XP...........P4
-------------------------------------------------------
<b>Absolute clock increase.............25.0%..........50.0%</b>
SiSoft CPU (Dhrystone):...............22.0%...........49.9%
SiSoft CPU (Whetstone):..............26.2%...........49.7%
SiSoft Multimedia (Integer):..........25.7%...........49.9%
SiSoft Multimedia (Floating-Point):.25.6%...........49.7%
-------------------------------------------------------
Lame MP3:.................................26.5%...........48.4%
Lightwave 7b:............................24.9%...........48.0%
In ALL other tests NONE of the two systems shows a linear relationship between performance and clockrate. Notice that the highly SSE/SSE2 optimized Lightwave 7b and Lame MP3 both lives by the clock.
In other applications both CPU’s are hold back by various bottlenecks in the their systems. The 512kb 2'nd level cache introduced in Northwood A and the upcoming Northwood B with 533MHz system bus (to be introduced later this spring) are both attempts to reduce the effects of the current external memory bottleneck.
7.Taynalator, if one reads all your posts in “P4 3010MHZ SUCKY!!!” thread, it becomes evident that you are nothing but a single-minded AMD zealot with totally unrealistic expectations in general system performance as the clockspeed goes up. Only in special applications where most of the job can be done within the CPU itself you will get a linear increase in performance as the clockspeed goes up. (Lame MP3 and Lightwave 7b are examples of this thanks to SSE/SSE2).
I actually don’t care if the P4 has to run at 50GHz in order to level with the Athlon XP. Likewise, I don’t care that AMD needs 3 FPU units and SSE to obtain a combined FPU performance that is still lower than what Intel can obtain with 1 FPU unit and SSE2. It’s the outcome that’s important. Look at the CPU as a black box, offering you performance. Why do you care about what’s inside? It doesn’t matter at all, but apparently you need something to point fingers at now, where your horse is struggling to catch up. What I care about is price, performance, stability and quality. For now, I think the P4/1800A or the P4/2000A together with a carefully selected motherboard is what fits the bill in all respects.
Face it; the once so embarrassing P4 has finally surpassed your precious little AMD counterpart. At least for a while you’ll have to live with it.
<i>/Copenhagen</i>
<b><i>Seagate Barracuda IV.
Bad performance in RAID setups!
</i></b>
I can't blame them for doing that, because the headline of my thred looked like a flame. The actual content however was no such thing. I'm just tired of headlines like "P4 sucks", "AMD Athlon XP sucks" and so on. Normally I don't bother, because it is evident that whoever is posting is doing so in order to start a flame war. With Taylanator however, I felt that he was being serios about his claims, as if he actually believed in the nonsense he spewed. Also I wanted to let him know that headlines like his "P4 3010MHZ SUCKY!!!" is annoying and stupid.
OK, but I still want my message through to get through to him, so here is a repost of my original posting:
@Taynalator:
As it is simply too overwhelming to respond to every little cry you’ve made over the last couple of days, I think I’ll just make some general comments and then take it from here. To be fair, I have to admit that it seems you are becoming more reasonable as time goes by. You have probably discovered that your bullish/stupid style won’t do you any good in making arguments. Nonetheless I still feel a need to set things straight.
1.The test was made primarily to show the overclockability of the Northwood. After the die-shrink it makes sense to examine its overclocking capabilities.
2.The other CPU’s listed in the benchmarks have just been included as a reference. They’ve not been made especially for this article but were all made some time ago. Do not regard this as an insult towards AMD (and you). It’s very time-consuming to do these kinds of tests, so for now the THG people have chosen to focus on the new 0.13 micron CPU offering from Intel. Probably later we will see something similar done for the AMD Athlon/Throughbred.
3.As I have already said some time ago, I’ve noticed that it seems to be very important for some test-sites to be quickly out with their reviews. That’s somewhat understandable, and the penalty seems to be bad editing at times. That’s not good if the article is trying to explain technically difficult stuff, but when it comes to benchmarks, it really doesn’t matter that much. If you use a little common sense you can draw your own conclusions in each benchmark. As long as they don’t fiddle with the results, I’m mostly happy.
4.There IS a problem with the German to English translation, but it is nothing major. I have visited the German site and found that the translation is actually adequate (being a Dane, I also “master” the German language), but the German site had two comments that are left out in the English translation. In the SiSoft Sandra benchmark the following comment can be found on the German site:
<i>”Der Pentium 4/3000 liegt bei allen Benchmarks vorn. Dennoch sind die Ergebnisse nur unter Vorbehalt zu verwenden. ”</i>
That could have been said a bit more precisely, but they are effectively saying that the synthetic memory benchmark is not reflecting the actual performance of a system. The translator, being a clever person, couldn’t match the results with the comment and probably simply left it out. Instead the comment should have been something like this: <i> The overclocked PC2100 DDR RAM on the P4/3000A system is not able to match the bandwidth of a standard RDRAM based system.</i>
The other comment can be found in the Cinema 4D bench:
<i>”Beim 3D-Rendering unter Cinema 4D zeigt sich, dass AMD und Intel sich in der Führungsposition abwechseln.”</i>
At best, this is a totally irrelevant comment, which the translator wisely left out. The P4/3000A is taking the lead in both tests with a significant margin. What the editor is maybe trying to say is that the blue(P4) and green(Athlon XP) bars interchange in the Shading (Cinema 4D) benchmark. Totally irrelevant.
5.As far as judging the performance of the P4 as “PITIFULL”, I think you are completely wrong. The P4/3000A displays an almost linear performance/clockspeed relationship in the SAME benchmarks as the AMD Athlon XP. In the following table, I’ve listed what a 25% increase in clockspeed does for the Athlon XP (1333MHz/XP1500+ to 1660MHz/XP2000+) and what a 50% increase in clockspeed does for the P4 (P4/2000A to P4/3000A).
Benchmark:..............................Athlon XP...........P4
-------------------------------------------------------
<b>Absolute clock increase.............25.0%..........50.0%</b>
SiSoft CPU (Dhrystone):...............22.0%...........49.9%
SiSoft CPU (Whetstone):..............26.2%...........49.7%
SiSoft Multimedia (Integer):..........25.7%...........49.9%
SiSoft Multimedia (Floating-Point):.25.6%...........49.7%
-------------------------------------------------------
Lame MP3:.................................26.5%...........48.4%
Lightwave 7b:............................24.9%...........48.0%
In ALL other tests NONE of the two systems shows a linear relationship between performance and clockrate. Notice that the highly SSE/SSE2 optimized Lightwave 7b and Lame MP3 both lives by the clock.
In other applications both CPU’s are hold back by various bottlenecks in the their systems. The 512kb 2'nd level cache introduced in Northwood A and the upcoming Northwood B with 533MHz system bus (to be introduced later this spring) are both attempts to reduce the effects of the current external memory bottleneck.
7.Taynalator, if one reads all your posts in “P4 3010MHZ SUCKY!!!” thread, it becomes evident that you are nothing but a single-minded AMD zealot with totally unrealistic expectations in general system performance as the clockspeed goes up. Only in special applications where most of the job can be done within the CPU itself you will get a linear increase in performance as the clockspeed goes up. (Lame MP3 and Lightwave 7b are examples of this thanks to SSE/SSE2).
I actually don’t care if the P4 has to run at 50GHz in order to level with the Athlon XP. Likewise, I don’t care that AMD needs 3 FPU units and SSE to obtain a combined FPU performance that is still lower than what Intel can obtain with 1 FPU unit and SSE2. It’s the outcome that’s important. Look at the CPU as a black box, offering you performance. Why do you care about what’s inside? It doesn’t matter at all, but apparently you need something to point fingers at now, where your horse is struggling to catch up. What I care about is price, performance, stability and quality. For now, I think the P4/1800A or the P4/2000A together with a carefully selected motherboard is what fits the bill in all respects.
Face it; the once so embarrassing P4 has finally surpassed your precious little AMD counterpart. At least for a while you’ll have to live with it.
<i>/Copenhagen</i>
<b><i>Seagate Barracuda IV.
Bad performance in RAID setups!
</i></b>